The stationary phase auxiliary field Monte Carlo method: a new strategy for reducing the sign problem of auxiliary field methodologies
In this paper we focus on a new computational procedure, which permits an efficient calculation within the classical auxiliary field methodology. As has been previously reported, the method suffers from a sign problem, typically encountered in methodologies based on a field-theoretical approach. To ameliorate its statistical convergence, the efforts have so far exclusively been concentrated on the development of efficient analytical integral transformation techniques, such as the method of Gaussian equivalent representation of Efimov et al. In the present work we reformulate the classical auxiliary field methodology according to the concepts of the stationary phase Monte Carlo method of Doll et al., a numerical strategy originally developed for the simulation with real-time path integrals. The procedure, which is here employed for the first time for auxiliary field computation, utilizes an importance sampling strategy, to identify the regions of configuration space that contribute most strongly to the functional integral averages. Its efficiency is here compared to the method of Gaussian equivalent representation.